On Mon, Dec 20, 2021 at 11:15:14AM -0800, Linus Torvalds wrote: > Well, that patch seems to be a no-op removal of dead code, so absolutely yes. > > That said, I think it would be good to split it up. I looked at that > patch and went "is that really a no-op" to the point of recreating it. > > I think it would be good to make it multiple patches that are each > individally trivial. IOW, start with > > (1) remove second argument to reuse_swap_page() that is always NULL, > without making any other changes > > (2) that now made 'total_mapcount' unused in reuse_swap_page(), > remove it as an argument from page_trans_huge_map_swapcount() > > (3) that now made 'total_mapcount' unused in > page_trans_huge_mapcount(), remove it as an argument there too. Hah, that was actually how I did it originally (without actually committing at each step, and with a few "Oh, hang on, now we can avoid calculating this too" stops and restarts along the way), but I thought it all hung together logically as a single change. It's hard to see things from the other person's perspective at times.
